IEEE PIC 2016 Keynote Speech (1): Signal processing techniques in medical ultrasound imaging

Abstract

Medical diagnostic ultrasound is widely used in clinical situations owing to its preferable properties, such as non-invasiveness and cost effectiveness. In addition, medical ultrasound imaging provides a significantly better temporal resolution than computed tomography and magnetic resonance imaging. Recently, the temporal resolution of ultrasound imaging has been further boosted to several thousand frames per second, and such an extremely high temporal resolution is predominantly useful for measurement of the cardiovascular dynamics. On the other hand, ultrasound imaging is a very flexible modality, which can control the tradeoff between the imaging temporal resolution and the spatial resolution (and contrast). In the current situation of medical ultrasonic imaging, the number of ultrasound transmissions per frame needs to be increased to obtain better spatial resolution and contrast, that degrades the temporal resolution. However, such a tradeoff problem can be solved by the signal processing technique. A key technology to realize high temporal and spatial resolution simultaneously is the adaptive beamforming technology, which realizes a significantly better spatial resolution than the non-adaptive conventional beamforming technology. However, the computational complexity of adaptive beamforming is an issue to be solved to deliver adaptive beamforming techniques in clinical situation. Furthermore, motion estimation techniques are very important to assess the tissue dynamics. However, current motion estimators require multiple ultrasound transmissions to obtain tissue motion and, therefore, the imaging frame rate is degraded. Also, in high-frame-rate ultrasound, several thousand frames per second need to be processed to obtain tissue motion. Therefore, a computationally efficient motion estimator needs to be developed. In this talk, I will present some solutions for such problems in high-frame-rate ultrasonic imaging. I will show that the computational complexity in adaptive beamforming can be reduced significantly without degradation of its performance. Also, I will show that accurate motion estimation is possible without increasing the number of ultrasound transmissions per frame by analyzing the frequency and phase of the received ultrasound echo, and potential applications of high-frame-rate ultrasound imaging will be shown.